Polyadenylation at RNA polymerase III loci underlies the accumulation of noncoding retrotransposons during herpesviral infection

The cellular accumulation of short non-coding RNAs (ncRNAs) transcribed by RNA Polymerase III (Pol III) is a hallmark of various cellular stressors and inflammatory-associated diseases. Yet, the mechanisms driving the accumulation of these RNAs are largely undefined. Infection with several DNA viruses is known to significantly alter the cellular Pol III transcriptome, leading to the induction of a class of non-coding retrotransposons known as short interspersed nuclear elements (SINE) and tRNA genes. Here, we sought to define the mechanisms driving Pol III transcribed ncRNA abundance during viral infection, using the murine herpesvirus MHV68 as a model. Our findings reveal that while the expression of Pol III transcripts, such as the murine-specific family of B2 SINE ncRNAs and pre-tRNAs, significantly increase during MHV68 infection, Pol III genomic occupancy is enhanced at a much fewer subset of B2 SINE and tRNA genes. Using DNA motif analyses and a convolutional neural network (CNN) based model, we identified non-promoter sequence elements within B2 SINE genes that distinguish infection-induced loci. We found that infection-induced B2 SINE genes are enriched for signal sequences that confer polyadenylation, and endogenous B2 SINE ncRNA polyadenylation depends on mRNA cleavage and polyadenylation (CPSF) machinery. We discovered that mRNA CPSF components are recruited to sites of Pol III transcription in response to MHV68 infection in a manner dependent on Pol III occupancy. Chromatin-associated B2 SINE ncRNAs are also bound by the CPSF complex, suggesting an RNA-dependent, co-transcriptional polyadenylation of B2 SINE ncRNAs. This uncovers an inducible, coupled relationship between Pol III transcription and mRNA-like polyadenylation of ncRNAs. Also, CPSF recruitment to Pol III genes is not restricted to murine genomes but also occurs at human SINE and tRNA genes, suggesting that this previously unknown coupled relationship may be a widespread feature of Pol III transcription.